Development of Quadrupole MassSpectrometers Using Rapid PrototypingTechnology

Boris Brkic,a Neil France,a Adam T. Clare,b Chris J. Sutcliffe,b

Paul R. Chalker,b and Stephen Taylora

a Department of Electrical Engineering and Electronics, University of Liverpool, Liverpool, United Kingdomb Department of Engineering, University of Liverpool, Liverpool, United Kingdom

In this report, we present a prototype design of a quadrupole mass filter (QMF) withhyperbolic electrodes, fabricated at the University of Liverpool using digital light processing(DLP), a low-cost and lightweight 3D rapid prototyping (RP) technique. Experimental massspectra are shown for H2

�, D2�, and He� ions to provide proof of principle that the DLP mass

filter is working as a mass analyzer in the low-mass range (1 to 10 amu). The performance ofthe DLP QMF has also been investigated for individual spectral peaks. Numericalsimulations of the instrument were performed by coupling CPO and Liverpool QMS-2programs to model both the ion source and mass filter, respectively, and the instrument isshown to perform as predicted by theory. DLP thus allows miniaturization of massspectrometers at low cost, using hyperbolic (or other) geometries of mass analyzerelectrodes that provide optimal ion manipulation and resolution for a given application.The potential of using RP fabrication techniques for developing miniature and microscalemass analyzers is also discussed. (J Am Soc Mass Spectrom 2009, 20, 1359 –1365) © 2009American Society for Mass Spectrometry

Development of mass spectrometers has ad-vanced rapidly in recent years with a focus onbuilding fully integrated devices that can be

made portable while maintaining good performance.To achieve this, miniaturization of mass spectrometercomponents has been pursued, especially in the case ofmass analyzers. This is because reduction in size ofanalyzers offers several advantages:

1. Lower manufacturing costs are possible because theimplementation technologies used often offer massproduction not only for separate components, butalso for complete devices.

2. Operation at higher pressures because of shorterlength of the ion mean free path.

3. Use of smaller and less expensive vacuum systemsattributed to smaller device size and higher pressureoperation.

4. Lower power consumption with the possibility ofoperation using low-power battery systems sincelower electrode voltages are needed to generate therequired electric fields.

5. Potential for the whole mass spectrometry system tobe made portable.

Address reprint requests to Dr. Boris Brkic, University of Liverpool,Department of Electrical Engineering and Electronics, Brownlow Hill,

Liverpool, Merseyside L69 3GJ, United Kingdom. E-mail: Boris.Brkic@liv.ac.uk

© 2009 American Society for Mass Spectrometry. Published by Elsevie1044-0305/09/$32.00doi:10.1016/j.jasms.2009.03.025

Therefore, these advantages will further increase thenumber of mass spectrometry applications, particularlyfor field applications such as instant medical diagnosis,water and environmental analysis, and detection of oil,natural gas, and explosives [1].

In the past, miniaturization of mass spectrometershad been done by using several different methods. Firstof all, there are semiconventional methods, which wereused for realization of miniature quadrupole massfilters (QMFs) and arrays of quadrupole analyzers forresidual gas analysis [2–5]. Another approach includesminiaturization of hyperbolic QMFs using ceramic ma-terial, coated with metal to define electrode regions [6,7]. In recent years, the most popular method for minia-turization of mass analyzers is microelectromechanicalsystems (MEMS) technology, which is mainly based onsemiconductor microfabrication processes. Since com-plex electrode geometries such as hyperbolic are diffi-cult to fabricate at the microscale using MEMS, simplergeometries such as cylindrical and planar are used togive approximation to ideal hyperbolic field. Miniaturemass analyzers constructed using MEMS include QMFswith cylindrical electrodes [8–10], time-of-flight massanalyzers [11], and several types of ion traps [12–15].Ion source MEMS miniaturization has also advancedwith realization of a carbon nanotube electron impaction source [16]. Finally, some rapid prototyping tech-niques have also been used for miniaturization of mass

analyzers, including an ultraviolet-LIGA process for

Published online April 5, 2009r Inc. Received December 10, 2008

Revised March 25, 2009Accepted March 26, 2009

1360 BRKIC ET AL. J Am Soc Mass Spectrom 2009, 20, 1359–1365

fabricating two-dimensional (2D) QMF arrays [17] andstereolithography (SLA) for fabricating rectilinear iontraps [18]. An attractive goal for most of the previouslymentioned miniaturization methods is a reliable andfully integrated mass spectrometer with all of its com-ponents built using the same process.

Here, we propose a novel approach for miniaturiza-tion of quadrupole mass spectrometers (QMS) usingrapid prototyping techniques. The technique used forfabrication of a hyperbolic QMF is digital light process-ing (DLP), which was initially invented by Texas Instru-ments (TI, Dallas, TX, USA) for video applications. DLPis based on a TI digital micromirror device (DMD),which is a MEMS semiconductor chip that containsmicroscopic mirrors aligned on a matrix. The purposeof micro mirrors is to enable precise control of the laserbeams to achieve high-resolution projection stereo-lithography for fabricating complex three-dimensional(3D) microstructures [19, 20]. DLP uses materials suchas polymethyl methacrylate (PMMA) to produce any3D shape with low fabrication errors at microscale. Wehave used DLP to fabricate PMMA hyperbolic elec-trodes for a QMF and housing for them. QMF elec-trodes were then coated with gold to provide electricalconductivity with low surface roughness for both thePMMA rods and the gold coating. The reason we choseto fabricate a QMF with hyperbolic rods is that it canprovide significantly better resolution at 10% of thepeak height than the QMF with circular rods. This hadbeen demonstrated experimentally by Brubaker [21],with an improvement of a factor of two in resolution, aswell as numerically by Gibson [22], who quantifiedimprovement in both resolution and transmission whenhyperbolic electrodes are used compared with circularelectrodes.

The following sections describe the methods used forDLP QMF fabrication, obtaining proof of principle formass spectrometry operation and its performance in-vestigation. Discussion follows of the advantages ofusing DLP for low-cost manufacturing not only forsingle mass analyzers with improved electromagneticfields, but also for other segmented structures such asion sources, prefiltered mass analyzers, and arrays ofanalyzers (e.g., trap arrays). Another advantage ofDLP—the potential for a fully integrated and light-weight miniature mass spectrometer—is also discussed.

Experimental

Modeling

Both analytical and numerical modeling of electrostaticfields are useful and often essential when designing ionsources and mass analyzers because they simulate theresults that appear in real systems. Our numericalmodel may be used to produce individual mass peaksor complete mass spectrum and it can support any typeof ion source together with QMF with electrodes of any

profile. The model works by coupling the commercially

available CPO3D program [23] with an in-house simu-lation program suite the QMS-2 [24].

CPO is an electrostatic simulation package basedon the boundary-element method (BEM), which hasproved to be more accurate than the finite-elementmethod (FEM) and the finite-difference method (FDM)for modeling electrostatic lenses [25]. It has also beenshown that CPO is more accurate than SIMION (FDM)for modeling miniature ion traps in free space [26]. Thisis mainly because the BEM uses only the surface of theelectrodes to define the grid points for calculatingpotentials, whereas the FEM and the FDM also use thespace enclosed by the electrodes. In this way, the BEMenables faster computation and higher accuracy evenwith a small number of electrode segments, which areused to adjust the simulation accuracy. Another advan-tage is that the number of segments can be defined fordifferent electrode regions and thus have larger num-bers of segments for most critical regions and smallernumbers of segments for regions where high accuracy isnot needed.

QMS-2 is a 2D simulation package for QMF, devel-oped by the University of Liverpool and was initiallybased on the FDM and, more recently, including theBEM for calculating electric fields and potentials. Itsupports QMFs with hyperbolic, cylindrical, and squareelectrodes, giving accurate performance predictions [22,24, 27], including the effect of electrode misalignment[28]. The purpose of the QMS-2 is to generate individualmass peaks and full mass spectra for given ion masseswithin a specified mass range for different stabilityzones. This is done by defining QMF dimensions anddrive parameters (voltages and frequency) as well asinitial ion oscillation parameters (positions, energies,and velocity components). The initial ion parameters inQMS-2 can be defined by setting constant ion energyangular spread in the direction at which ions enter theQMF. Alternatively, a custom approach can be used bymodeling ion motion within a given ion source usingprograms such as CPO and SIMION to obtain entranceion parameters. This has a more accurate correspon-dence to a real system, especially for ion energies thatare directly dependent on the voltages applied to theion source lenses.

For this work, CPO was used to simulate ion trajec-tories for the electron impact ion source (EIIS) andQMS-2 was used for the hyperbolic QMF that wasfabricated using rapid prototyping DLP. A hyperbolicQMF simulation was also added to CPO simulation ofthe EIIS to include the effect of the fringing fields as ionsenter the mass filter. The effect of space charge was alsosimulated between ions within the ion source.

Design

Before fabrication of the hyperbolic QMF, a detailedcomputer-aided design (CAD) drawing was done inPro/ENGINEER [29] to define the dimensions for every

component. Such CAD drawings are directly loaded

1361J Am Soc Mass Spectrom 2009, 20, 1359–1365 QMS IMPLEMENTATION WITH RP TECHNIQUES

into the DLP system (described in the following text),which then manufactures desired 3D shapes accordingto the Pro/ENGINEER design. Figure 1a and b showthe design drawing for the assembly of the hyperbolicQMF including electrodes, electrode housing, and pins.The shape of the grooves on the housing was carefullychosen to ensure tight fitting for the rods to establish agood alignment and uniform separation. This is espe-cially important for QMFs where even a small displace-ment of the electrodes along the y-axis can severelyreduce the performance of the instrument [28]. The pinswere designed to enable electrical connection to theelectrodes through the tiny holes and to further securethe electrodes within the housing.

Fabrication Technique

As mentioned previously, fabrication of the hyperbolicQMF was done using a DLP lithographic technique,based on the DMD. DLP has a dynamic maskingcapability to selectively expose photosensitive, resinousmaterials such as PMMA, causing them to polymerizelayer by layer to realize a desired geometry. Unlikeconventional lithographic techniques that require a spe-cific mask per given pattern layer, at a significant cost in

Figure 1. Pro/ENGINEER CAD design drawings for the assem-bly of the hyperbolic DLP QMF: showing QMF electrode design

(a) and design for electrode grooves on the housing (b).

most cases, with the DLP technique masking for eachgiven layer is simply reprogrammed. In this way, theDLP uses a dynamic masking regime to induce re-peated layers of a controlled geometry layer by layer.

Variations on the DLP theme are commonplacewithin the rapid prototyping (RP) market, althoughmost systems consist of the following: DMD chassis(plus optics), stepper motor (for z-direction transla-tions), resin container (with transparent floor), and aremovable built substrate. The preprocessing of thecomputer-generated model for DLP is identical to thatof the generic RP model, which makes it well suitablefor mass production. The resolution of DLP systems isanisotropic. This is because resolution-limiting param-eters are different in the coordinate axis with referenceto each built layer. Geometries realized in this way canbe thought of as consisting of a multitude of voxels(volumetric pixels). These discrete “building blocks”have their dimensions forced on them by the dimen-sions of each mirror mounted on the DMD (afterfocusing as observed at the resin floor), correspondingto the x–y resolution and in the z-direction the smallestpossible translation of the stepper motor. For our DLPstructures, the x–y voxel resolution used was 20 �m andin the z-direction, 25 �m.

Good tolerances/surface finish, minimal secondaryprocessing, and quick component interassembly are alladvantages of DLP. The relative juvenility of DLP as amanufacturing technique means that its full potential isnot yet realized and that considerable research needs tobe taken within various institutions and companies. Forour DLP QMF, a commercial envisionTEC PerfactorySystem (Gladbeck, Germany) was used to fabricatepins, electrodes, and the housing for them. Gold coatingfor the PMMA pins and electrodes was done by thermalevaporation with an Edwards E306A Coating System(Wilmington, MA, USA).

Experimental Setup

The fully assembled DLP QMF was coupled to acustom-built electronic control unit (ECU) and tested ina miniature vacuum system. The ECU was used tocontrol the ion source voltages, to provide radiofre-quency (rf) and dc drive voltages for the mass filter andto measure the current in the Faraday cup detector. Itwas linked via USB to a laptop PC for display of themass spectra obtained. The vacuum system consisted ofan Edwards two-stage rotary pump and an Edwardsturbomolecular pump (for pressures down to 5 � 10�6

Torr) with total pressure monitored using a vacuumgauge (Leybold Ionivac GmbH, Cologne, Germany).

Results and Discussion

The hyperbolic QMF, fabricated using DLP, was de-signed to fit into our existing vacuum flange and metal

housing as well as with our existing EIIS and detector.

1362 BRKIC ET AL. J Am Soc Mass Spectrom 2009, 20, 1359–1365

The DLP QMF has r0 � 2 mm, where r0 is half of theshortest distance between the opposing electrodes ofthe QMF. Length of the electrodes tested was 50 mm. Acommercial EIIS, built with conventional engineering,was used to test the DLP QMF. The EIIS consists of acage where ions are generated, an extracting lens at-tached to the cage to release ions, a focusing lens tofocus ions toward the QMF, and a decelerating lensheld at ground to reduce ion energies before they enterthe QMF. The cage has a cylindrical shape with diam-eter of 6 mm and length of 10 mm. All three lenses have0.3-mm thickness and 0.8-mm separation betweenthem, where the decelerating lens is separated from theQMF by 0.5 mm. All the lenses have re � 1.5 mm, wherere is the exit aperture radius.

Figure 2a shows uncoated and coated DLP QMFrods after fabrication. The thickness of the gold coatingwas initially about 1 �m, which proved to be suffi-ciently high to allow QMF operation. The resistance ofthe conducting electrodes from one end to another ininitial designs was roughly 40 �. By increasing thethickness of the gold coating, the resistance along theelectrodes was reduced to only 0.1 �, which gave more

Figure 2. Gold-coated and non-coated electrodes of the DLPQMF (a) and microscopic image of the non-coated PMMA rod to

show voxels and build direction (b).

accurate driving voltages and enabled better optimiza-tion of spectral peaks. We did notice that after contin-uous use, the gold coating can wear slightly from thePMMA. To overcome this problem in the future, wepropose to coat the electrodes with titanium, and thenwith gold, which should provide a better interface withPMMA. Figure 2b shows a picture of the enlarged partof the noncoated PMMA rod, taken by scanningelectron microscope. The surface roughness of theDLP rods was close to 1.5 �m, measured using aWyko NT 3300 profiling system (white light inter-ferometry; Veeco Instruments, Inc., Plainview, NY,USA). The roughness of electrode surfaces is espe-cially important for miniature mass analyzers, wherepatch potentials can increase ion motional heating ifelectrodes are not sufficiently smooth, thus causingthe distortion of ion motion [30].

Figure 3a shows the cross section of the assembly ofthe DLP QMF with r0 � 2 mm. The unmetallizedPMMA material of the housing behaves as a goodinsulator. Grooves within the housing were accuratelyfabricated so that the precision of electrode alignment isrelatively high with around 1% error. Figure 3b showsthe enlarged image of the cross section of the hyperbolicDLP QMF, confirming good electrode alignment at theends.

For initial QMS testing and for showing proof ofprinciple, the QMF was driven at a frequency of 3.686MHz. Ion source cage and the extracting lens were heldat 3 V, the focusing lens was held at �39 V, and thedecelerating lens was at 0 V. The emission current forionization was 0.8 mA and operating pressure was1.2 � 10�4 Torr. Figure 4a shows an experimental massspectrum for a H2/D2/He equal-concentration gas mix-ture measured using a commercial QMS with circularelectrodes having prefilter and postfilters, manufac-tured by MKS Instruments (MKS Cirrus, Andover, MA,USA). The MKS QMS has r0 � 3.175, electrode length of100 mm, and is driven at 1.843 MHz. The MKS QMS istuned to about 1-amu resolution at 10% of the peakheight across the operating mass range. The resolutionfor H2

� ions at mass 2 is 3.17 and the resolution for D2�

and He� ions, which are not resolved at mass 4, is 5.88at 10% of the peak height. Figure 4b shows a completeexperimental mass spectrum for the same H2/D2/Hegas mixture obtained from the DLP QMS. The resolu-tion for H2

� ions at mass 2 is 3.03 and the resolution forD2

� and He� ions at mass 4 is 5.13 at 10% of the peakheight. A very good correlation between the ratios ofspectral peaks can be seen by comparing the massspectra from the commercial MKS QMS with the DLPQMS. This provides proof of principle of the DLP QMSinstrument in the technologically important 1 to 10amu mass range. The ECU used was originally de-signed for microscale mass filters (submillimeter r0) andprovides only low-output voltages and reduced massrange. By reducing the size of r0 in future instrumentrealization, the mass range of the DLP instrument can

be extended using the existing ECU.

1363J Am Soc Mass Spectrom 2009, 20, 1359–1365 QMS IMPLEMENTATION WITH RP TECHNIQUES

To assess performance of the DLP QMF, optimiza-tions of spectral peaks at mass 2 and 4, respectively,were obtained by altering the extracting voltage on theion source cage (ion energy) and by changing the U/Vratio. For optimization of a mass 2 peak, the cagevoltage was varied in five steps with values of 4, 3, 2.5,2, and 1.5 V, where the lowest voltage provides thelowest ion energy and highest resolution. The massanalyzer was driven at 3.686 MHz. Figure 5a shows theoptimal simulated mass peak for 2H� ions within thehyperbolic DLP QMF operating in zone 1 stabilityregion with 1.5 V on the cage, obtained using CPO andQMS-2 programs. The simulated instrument resolutionat 10% of the peak height is 33.33. Figure 5b shows thecorresponding experimental mass peak for 2H� ionswith resolution of 30.5 at 10% of the peak height, whichis almost as high as the theoretical prediction. Compar-ing Figure 5a and b, a close similarity can be seenbetween the shapes of the simulated and experimen-tal mass peaks for m/z � 2 as well as low-iontransmission ascribed to low ion energies. Figure 6ashows the optimization steps for increased resolution

Figure 3. Cross sections of the DLP QMF prototype: showing theassembly (a) and electrode alignment (b).

mass 2 peaks obtained by reducing the ion source cage

extraction voltage. Figure 6b shows variation of ionpeak current with instrument resolution for mass 2H�

ions for the given optimization steps. Figure 6c showsvariation of ion peak current with instrument resolutionfor m/z � 4 by varying U/V ratio in five steps withvalues of 0.158, 0.160, 0.162, 0.164, and 0.166. The cageextraction voltage was set to 1.2 V and the massanalyzer was driven at 2 MHz to obtain mass 4 peaks.The highest resolution of 69.45 at 10% of the peak heightwas achieved with the highest U/V ratio that corre-sponds to the peak of the zone 1 stability diagram,resulting in the lowest ion peak current.

We have also investigated outgassing of the PMMAmaterial and noticed the increase from our system basepressure of 5 � 10�6 Torr with an empty vacuumsystem to 2 � 10�5 Torr with the DLP QMF inside, butwe did not see any negative effects to the measuredmass spectra and performance of the instrument in thismass range. Even though the DLP QMF was positionednear the electron impact ion source, we did not see anydamage to the electrodes and assembly housing.

Apart from the ability to reproduce low-cost minia-ture structures with good tolerances and smooth sur-faces, another major advantage of the DLP over othersimilar technologies is that it can be used for buildingsegmented structures. For example, ion source lensescan easily be made with DLP, where the lens and theinterlens insulator can be built as one polymer part,making a significant simplification. In this way, by

Figure 4. Experimental mass spectra for H2/D2/He gas mixture:showing spectrum obtained from the commercial MKS QMS withcircular electrodes (a) and spectrum obtained from the hyperbolic

DLP QMF (b).

1364 BRKIC ET AL. J Am Soc Mass Spectrom 2009, 20, 1359–1365

using a suitable mask, only segments that need to beconductive can be coated with gold (e.g., exit hole),while leaving other parts insulated. Another example ofusing DLP for segmented systems is for quadrupolemass filters that use prefilters for better sensitivity andpostfilters for better detection. In this case, the elec-trodes of prefilters, main analyzer, and postfilterswould all be part of the same PMMA rod, which wouldbe coated accordingly to define conducting regions foreach of these devices. In the same way, linear ion trapsor trap arrays can be implemented in DLP so that rf anddc electrodes are all part of the same polymer structure.DLP structures with multiple parts will be significantlylighter than their conventional counterparts made withmetal and ceramic insulators. This is advantageous forapplications where light weight is important, such asspacecraft.

Conclusions

A hyperbolic quadrupole mass filter has been fabricatedusing a digital light processing technique and assem-bled with a conventional electron impact ion source andFaraday detector into a quadrupole mass spectrometer(QMS). Experimentally obtained mass spectra for aH2/D2/He gas mixture was obtained and found toagree both quantitatively and qualitatively with spectraobtained using a commercial triple-filter QMS for the

Figure 5. Mass peaks for 2H� ions for the hyperbolic DLP QMF:showing optimal simulated peak, generated by coupling CPO andLiverpool QMS-2 programs (a) and optimal experimental peak,obtained by reduction of the ion source extraction voltage (b).

same gas mixture, proving the principle of DLP QMS

operation. QMS performance was obtained as expectedfrom full simulation of the instrument, including ionsource effects using a combination of commercial andin-house simulation software.

The DLP technique was found to be particularlysuitable for implementing mass analyzers that requireideal hyperbolic fields or for any other devices withcomplex geometries. Compared to other rapid proto-typing techniques, such as selective laser melting andselective laser ablation, DLP offers significantly smallerfeature size that allows a higher degree of accuracy. Italso has the potential to produce fully integrated massspectrometers with a single mass analyzer or with anarray of analyzers. Future work will focus on realizationof mass analyzers and other mass spectrometer compo-nents at the submillimeter and micro scales.

Figure 6. Experimental optimizations of the hyperbolic DLPQMF spectral peaks: optimization of mass 2 peaks by altering theion source cage extraction voltage (a), and experimental variationof ion peak current with instrument resolution for mass 2 (b) and

for mass 4 (c).

1365J Am Soc Mass Spectrom 2009, 20, 1359–1365 QMS IMPLEMENTATION WITH RP TECHNIQUES

AcknowledgmentsWe gratefully acknowledge EPSRC (UK) for funding this workunder the Follow on Fund programme.

References1. Taylor, S.; Bierbaum, V. M. Focus on Harsh Environment Mass Spec-

trometry. J. Am. Soc. Mass Spectrom. 2008, 19, 1375–1376.2. Ferran, R. J.; Boumsellek, S. High-Pressure Effects in Miniature Arrays

of Quadrupole Analyzers for Residual Gas Analysis from 10�9 to 10�2

Torr. J. Vac. Sci. Technol. A 1996, 14, 1258–1265.3. Holkeboer, D. H.; Karandy, T. L.; Currier, F. C.; Frees, L. C.; Ellefson,

R. E. Miniature Quadrupole Residual Gas Analyzer for Process Moni-toring at milliTorr Pressures. J. Vac. Sci. Technol. A 1998, 16, 1157–1162.

4. Boumsellek, S.; Ferran, R. J. Miniature Quadrupole Arrays for Residualand Process Gas Analysis. J. Inst. Environ. Sci. Technol. 1999, 42, 27–31.

5. Ketola, R. A.; Kiuru, J. T.; Tarkiainen, V.; Kotiaho, T.; Sysoev, A. A.Comparison of Analytical Performances of a Micro-Array QuadrupoleInstrument and a Conventional Quadrupole Mass SpectrometerEquipped with Membrane Inlets. Rapid Commun. Mass Spectrom. 2003,17, 753–756.

6. Hiroki, S.; Abe, T.; Murakami, Y.; Takano, Y.; Higuchi, M.; Miyake, M.Development of a QMS with a Ceramic Single-Piece Quadrupole.Vacuum 1993, 44, 71–74.

7. Wang, J.; Zhang, X.; Mao, F.; Xiao, M.; Cui, Y.; den Engelsen, D.; Lei, W.Study of a Micro Chamber Quadrupole Mass Spectrometer. J. Vac. Sci.Technol. A 2008, 26, 239–243.

8. Taylor, S.; Tindall, R. F.; Syms, R. R. A. Silicon-Based Quadrupole MassSpectrometry Using Microelectromechanical Systems. J. Vac. Sci. Tech-nol. B 2001, 19, 557–562.

9. Geear, M.; Syms, R. R. A.; Wright, S.; Holmes, A. S. Monolithic MEMSQuadrupole Mass Spectrometers by Deep Silicon Etching. IEEE/ASMEJ. Microelectromech. Syst. 2005, 14, 1156–1166.

10. Velasquez-Garcia, L. F.; Akinwande, A. I. An Out-of-Plane MEMSQuadrupole for a Portable Mass Spectrometer. Proc. Transducers Conf.2007, 2315–2320.

11. Wapelhorst, E.; Hauschild, J. P.; Müller, J. Complex MEMS: A FullyIntegrated TOF Micro Mass Spectrometer. Sens. Actuators A 2007, 138,22–27.

12. Blain, M. G.; Riter, L. S.; Cruz, D.; Austin, D. E.; Wu, G.; Plass, W. R.;Cooks, R. G. Towards the Hand-Held Mass Spectrometer: DesignConsiderations, Simulation, and Fabrication of Micrometer-Scaled Cy-lindrical Ion Traps. Int. J. Mass Spectrom. 2004, 236, 91–104.

13. Pau, S.; Pai, C. S.; Low, Y. L.; Moxom, J.; Reilly, P. T. A.; Whitten, W. B.;

Ramsey, J. M. Microfabricated Quadrupole Ion Trap for Mass Spectrom-eter Applications. Phys. Rev. Lett. 2006, 96, 120801.

14. Van Amerom, F. H. W.; Chaudhary, A.; Cardenas, M.; Bumgarner, J.;Short, R. T. Microfabrication of Cylindrical Ion Trap Mass SpectrometerArrays for Handheld Chemical Analyzers. Chem. Eng. Comm. 2008, 195,98–114.

15. Austin, D. E.; Wang, M.; Tolley, S. E.; Maas, J. D.; Hawkins, A. R.;Rockwood, A. L.; Tolley, H. D.; Lee, E. D.; Lee. M. L. Halo Ion Trap MassSpectrometer. Anal. Chem. 2007, 79, 2927–2932.

16. Bower, C. A.; Gilchrist, K. H.; Piascik, J. R.; Stoner, B. R.; Natarajan, S.;Parker, C. B.; Wolter, S. D.; Glass, J. T. On-Chip Electron Impact IonSource Using Carbon Nanotube Field Emitters. Appl. Phys. Lett. 2007, 90,124102.

17. Wiberg, D. V.; Myung, N. V.; Eyre, B.; Shcheglov, K.; Orient, O. J.;Moore, E.; Munz, P. LIGA-Fabricated Two-Dimensional QuadrupoleArray and Scroll Pump for Miniature Gas Chromatograph/Mass Spec-trometer. Proc. SPIE 2003, 4878, 8–13.

18. Fico, M.; Yu, M.; Ouyang, Z.; Cooks, R. G.; Chappell, W. J. Miniatur-ization and Geometry Optimization of a Polymer-Based Rectilinear IonTrap. Anal. Chem. 2007, 79, 8076–8082.

19. Takahashi, T.; Akahane, T.; High Speed Rapid Prototyping using aDigital Micromirror Device. Tech. Digest 17th Sensor Symposium 2000,271–274.

20. Sun, C.; Fang, N.; Wu, D. M.; Zhang, X. Projection Micro-StereolithographyUsing Digital Micro-Mirror Dynamic Mask. Sens. Actuators A 2004, 121,113–120.

21. Brubaker, W. M. NASA Report. NASW 1970, 1298.22. Gibson, J. R.; Taylor, S. Prediction of Quadrupole Mass Filter Perfor-

mance for Hyperbolic and Circular Cross Section Electrodes. RapidCommun. Mass Spectrom. 2000, 14, 1669–1673.

23. CPO programs, available at www.electronoptics.com.24. Hogan, T. J.; Taylor, S. Performance Simulation of a Quadrupole Mass

Filter Operating in the First and Third Stability Zones. IEEE Trans. Instr.Meas. 2008, 57, 498–508.

25. Cubric, D.; Lencova, B.; Read, F. H.; Zlamal, J. Comparison of FDM,FEM and BEM for Electrostatic Charged Particle Optics. Nucl. Instr.Methods Phys. Res. A 1999, 427, 357–362.

26. Brkic, B.; Taylor, S.; Ralph, J. F.; France, N. High-Fidelity Simulations ofIon Trajectories in Miniature Ion Traps Using the Boundary-ElementMethod. Phys. Rev. A 2006, 73, 012326.

27. Gibson, J. R.; Taylor, S. Asymmetrical Features of Mass Spectral PeaksProduced by Quadrupole Mass Filters. Rapid Commun. Mass Spectrom.2003, 17, 1051–1055.

28. Taylor, S; Gibson, J. R. Prediction of the Effects of Imperfect Construc-tion of a QMS Filter. J. Mass Spectrom. 2008, 43, 609–616.

29. Pro/ENGINEER program, available at www.ptc.com.30. Turchette, Q. A.; Kielpinski, D.; King, B. E.; Leibfried, D.; Meekhof,

D. M.; Myatt, C. J.; Rowe, M. A.; Sackett, C. A.; Wood, C. S.; Itano,

W. M.; Monroe, C.; Wineland, D. J. Heating of Trapped Ions from theQuantum Ground State. Phys. Rev. A 2000, 61, 063418.